Microsoft MDASH AI scanner boosts vulnerability detection to 96.5%

Microsoft introduced codename MDASH, a multi-model AI-powered scanning system for detecting and remediating software vulnerabilities. Unlike traditional methods that rely on fixed security reviews, MDASH operates at enterprise scale by integrating directly with development tools like GitHub and Azure DevOps. The system, recently implemented in Microsoft engineering teams, improves vulnerability hunting across complex surfaces, including Windows and Azure infrastructure.

Since its launch, MDASH achieved a 96.5% success rate in identifying vulnerabilities, highlighting key improvements in scoping, threat modeling, and reasoning capabilities. The ongoing advancements aim to enhance the system's real-world effectiveness and speed up the process of securing code against emerging threats.